From 61fe39a4882b6b3a4a047fc6c56f6cd173a5da5a Mon Sep 17 00:00:00 2001
From: Sawyer Cutler <sawyer@dirtroad.dev>
Date: Sun, 20 Sep 2026 21:03:02 -0700
Subject: [PATCH 1/6] fix(prompts): budget greybeard-grok overage from ceremony
 fold-in + #1121 growth

---
 src/agent/prompt-sizes.test.ts | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)

diff --git a/src/agent/prompt-sizes.test.ts b/src/agent/prompt-sizes.test.ts
index 719e336ef..dc3136cf4 100644
--- a/src/agent/prompt-sizes.test.ts
+++ b/src/agent/prompt-sizes.test.ts
@@ -76,14 +76,18 @@ const PROMPT_SIZE_BASELINE: Record<
 
 /**
  * Deliberate budgets above baseline + allowance, with justification.
- * Empty after the origin/main rebase: every main budget fits within fresh
- * baseline + allowance (draper/warden included), and the entries where main
- * reads higher (intern, testsmith, gauntlet, prober) are stale-measurement
- * residue, not deliberate over-allowance.
+ * greybeard: the grok residual (tool budget + 8-line ceremony, folded into the
+ * canonical promptResidual seam verbatim under CL-8296) plus the upstream
+ * greybeard-package growth (#1121) pushed greybeard-grok to 8218 chars,
+ * 218 over the 8000 baseline + allowance budget. Trimming the greybeard body
+ * is greybeard-lane-owned, so the overage is budgeted here instead; bytes
+ * stay at the current budget level (measured 8248 < 9000).
  */
 const PROMPT_SIZE_OVERRIDES: Partial<
   Record<DirectorId, { chars: number; bytes: number }>
-> = {};
+> = {
+  greybeard: { chars: 8300, bytes: 9000 },
+};
 
 const CHAR_ALLOWANCE = 2000;
 const BYTE_ALLOWANCE = 3000;

From 72a264b4509e90e9df8c2e76d54e9d6ed086c912 Mon Sep 17 00:00:00 2001
From: Sawyer <sawyer@dirtroad.dev>
Date: Fri, 18 Sep 2026 15:16:11 -0700
Subject: [PATCH 2/6] =?UTF-8?q?test(prompt-variance):=20red=20=E2=80=94=20?=
 =?UTF-8?q?failing=20tests=20for=20family=20rows,=20assemble,=20per-row=20?=
 =?UTF-8?q?sizes?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

CL-8269: versioned prompt-variance package (default/muse/grok rows,
assemble, per-model overrides, per-row sizes) plus three-family
director prompt-size coverage. All fail until the package lands.
---
 packages/prompt-variance/src/assemble.test.ts | 52 +++++++++++++
 packages/prompt-variance/src/resolve.test.ts  | 59 ++++++++++++++
 packages/prompt-variance/src/rows.test.ts     | 78 +++++++++++++++++++
 packages/prompt-variance/src/sizes.test.ts    | 24 ++++++
 src/agent/prompt-sizes.test.ts                | 21 ++++-
 5 files changed, 231 insertions(+), 3 deletions(-)
 create mode 100644 packages/prompt-variance/src/assemble.test.ts
 create mode 100644 packages/prompt-variance/src/resolve.test.ts
 create mode 100644 packages/prompt-variance/src/rows.test.ts
 create mode 100644 packages/prompt-variance/src/sizes.test.ts

diff --git a/packages/prompt-variance/src/assemble.test.ts b/packages/prompt-variance/src/assemble.test.ts
new file mode 100644
index 000000000..a7e2b3ec8
--- /dev/null
+++ b/packages/prompt-variance/src/assemble.test.ts
@@ -0,0 +1,52 @@
+import { describe, expect, test } from "bun:test";
+import { assemble } from "./assemble.js";
+import { defaultRow, grokRow, museRow } from "./rows.js";
+
+// CL-8269 RED: assemble does not exist yet — every test below fails until
+// the GREEN lands packages/prompt-variance.
+
+const SECTIONS = ["contract", "Tools (names only): read_file", "context"];
+const TOOLS = ["read_file", "skill_search", "use_skill", "run_shell"];
+
+describe("assemble", () => {
+  test("joins sections with the residual as the last section", () => {
+    const out = assemble(SECTIONS, grokRow, TOOLS);
+    expect(out.systemPrompt).toContain("contract");
+    expect(out.systemPrompt.trimEnd().endsWith(grokRow.residual)).toBe(true);
+  });
+
+  test("a row without residual leaves the sections unchanged", () => {
+    const out = assemble(SECTIONS, defaultRow, TOOLS);
+    expect(out.systemPrompt).toBe(SECTIONS.join("\n\n"));
+  });
+
+  test("muse residual closes the prompt like the shipped tail append", () => {
+    const out = assemble(SECTIONS, museRow, TOOLS);
+    expect(out.systemPrompt.trimEnd().endsWith(museRow.residual)).toBe(true);
+  });
+
+  test("filters advertisedToolDeny from the tool names, preserving order", () => {
+    const out = assemble(SECTIONS, grokRow, TOOLS);
+    expect([...out.toolNames]).toEqual(["read_file", "use_skill", "run_shell"]);
+  });
+
+  test("variance cannot grant tools: output is always a subset of the input", () => {
+    for (const row of [defaultRow, museRow, grokRow]) {
+      const out = assemble(SECTIONS, row, TOOLS);
+      for (const name of out.toolNames) {
+        expect(TOOLS).toContain(name);
+      }
+      expect(out.toolNames.length).toBeLessThanOrEqual(TOOLS.length);
+    }
+  });
+
+  test("variance cannot grant tools: unknown mounted names pass through untouched", () => {
+    const out = assemble(SECTIONS, defaultRow, ["alpha", "beta"]);
+    expect([...out.toolNames]).toEqual(["alpha", "beta"]);
+  });
+
+  test("refuses a row that denies use_skill", () => {
+    const bad = { ...grokRow, advertisedToolDeny: ["use_skill"] };
+    expect(() => assemble(SECTIONS, bad, TOOLS)).toThrow(/use_skill/);
+  });
+});
diff --git a/packages/prompt-variance/src/resolve.test.ts b/packages/prompt-variance/src/resolve.test.ts
new file mode 100644
index 000000000..f98786ed5
--- /dev/null
+++ b/packages/prompt-variance/src/resolve.test.ts
@@ -0,0 +1,59 @@
+import { describe, expect, test } from "bun:test";
+import { applyRowOverride, resolvePromptVariance } from "./resolve.js";
+import { grokRow } from "./rows.js";
+
+// CL-8269 RED: the resolver does not exist yet — every test below fails
+// until the GREEN lands packages/prompt-variance.
+
+describe("resolvePromptVariance", () => {
+  test("resolves each family to its row", () => {
+    expect(resolvePromptVariance({ family: "default" }).id).toBe("default");
+    expect(resolvePromptVariance({ family: "muse" }).id).toBe("muse");
+    expect(resolvePromptVariance({ family: "grok" }).id).toBe("grok");
+  });
+
+  test("grok keeps its finish-bias residual and skill_search deny on leaves", () => {
+    const row = resolvePromptVariance({ family: "grok", orchestrator: false });
+    expect(row.residual).toBe(grokRow.residual);
+    expect(row.advertisedToolDeny).toContain("skill_search");
+  });
+
+  test("grok on orchestrators falls back to the default shape", () => {
+    const row = resolvePromptVariance({ family: "grok", orchestrator: true });
+    expect(row.residual).toBe("");
+    expect([...row.advertisedToolDeny]).toEqual([]);
+  });
+
+  test("muse keeps its residual on leaves and orchestrators alike", () => {
+    expect(
+      resolvePromptVariance({ family: "muse", orchestrator: false }).residual,
+    ).not.toBe("");
+    expect(
+      resolvePromptVariance({ family: "muse", orchestrator: true }).residual,
+    ).not.toBe("");
+  });
+
+  test("applies per-model-id overrides from the row", () => {
+    const row = applyRowOverride(
+      {
+        ...grokRow,
+        overrides: {
+          "grok-4-special": { advertisedToolDeny: [] },
+        },
+      },
+      "GROK-4-SPECIAL",
+    );
+    expect([...row.advertisedToolDeny]).toEqual([]);
+    expect(row.residual).toBe(grokRow.residual);
+  });
+
+  test("an unknown model id resolves to the base row", () => {
+    const row = resolvePromptVariance({
+      family: "grok",
+      orchestrator: false,
+      model: "grok-9-unknown",
+    });
+    expect(row.residual).toBe(grokRow.residual);
+    expect([...row.advertisedToolDeny]).toEqual(["skill_search"]);
+  });
+});
diff --git a/packages/prompt-variance/src/rows.test.ts b/packages/prompt-variance/src/rows.test.ts
new file mode 100644
index 000000000..18441687d
--- /dev/null
+++ b/packages/prompt-variance/src/rows.test.ts
@@ -0,0 +1,78 @@
+import { describe, expect, test } from "bun:test";
+import {
+  defaultRow,
+  FAMILY_IDS,
+  grokRow,
+  museRow,
+  type PromptVarianceFamily,
+} from "./rows.js";
+
+// CL-8269 RED: the versioned prompt-variance package does not exist yet, so
+// every test below fails at import time until the GREEN lands it.
+
+describe("prompt-variance family rows", () => {
+  test("ships exactly the default/muse/grok families", () => {
+    expect([...FAMILY_IDS]).toEqual(["default", "muse", "grok"]);
+  });
+
+  test("has no glm/claude/gpt rows until their evals land (CL-8265/7772/7775)", () => {
+    for (const id of FAMILY_IDS) {
+      expect(["glm", "claude", "gpt"]).not.toContain(id);
+    }
+  });
+
+  test("every row carries the render/residual/deny/omit shape", () => {
+    for (const row of [defaultRow, museRow, grokRow]) {
+      expect(row.render).toBe("tail");
+      expect(typeof row.residual).toBe("string");
+      expect(Array.isArray([...row.advertisedToolDeny])).toBe(true);
+      expect(Array.isArray([...row.sectionOmit])).toBe(true);
+    }
+  });
+
+  test("row ids match their family", () => {
+    const ids: PromptVarianceFamily[] = [
+      defaultRow.id,
+      museRow.id,
+      grokRow.id,
+    ];
+    expect(ids).toEqual(["default", "muse", "grok"]);
+  });
+
+  test("muse row is the shipped CL-7869 tool-discipline text", () => {
+    expect(museRow.residual).toContain("Tool discipline:");
+    expect(museRow.residual).toContain("Batch independent tool calls");
+    expect(museRow.residual).toContain("Never re-read a file");
+    expect(museRow.residual).toContain("Do not narrate; act.");
+  });
+
+  test("grok row is the existing finish-bias residual", () => {
+    expect(grokRow.residual).toContain("Finish bias (xAI / Grok worker):");
+    expect(grokRow.residual).toContain("prefer the structured report");
+    expect(grokRow.residual).toContain("re-open paths you already read");
+    expect(grokRow.residual).toContain("done-definition is met");
+    expect(grokRow.residual).toContain("never run_shell");
+  });
+
+  test("default row carries no residual", () => {
+    expect(defaultRow.residual).toBe("");
+  });
+
+  test("grok leaves deny skill_search only; other rows deny nothing", () => {
+    expect([...grokRow.advertisedToolDeny]).toEqual(["skill_search"]);
+    expect([...defaultRow.advertisedToolDeny]).toEqual([]);
+    expect([...museRow.advertisedToolDeny]).toEqual([]);
+  });
+
+  test("no row denies use_skill — brief-named skills always load directly", () => {
+    for (const row of [defaultRow, museRow, grokRow]) {
+      expect(row.advertisedToolDeny).not.toContain("use_skill");
+    }
+  });
+
+  test("sectionOmit starts empty on every row", () => {
+    for (const row of [defaultRow, museRow, grokRow]) {
+      expect([...row.sectionOmit]).toEqual([]);
+    }
+  });
+});
diff --git a/packages/prompt-variance/src/sizes.test.ts b/packages/prompt-variance/src/sizes.test.ts
new file mode 100644
index 000000000..66e2c618a
--- /dev/null
+++ b/packages/prompt-variance/src/sizes.test.ts
@@ -0,0 +1,24 @@
+import { describe, expect, test } from "bun:test";
+import { defaultRow, grokRow, museRow } from "./rows.js";
+
+// CL-8269 RED: per-family-row prompt sizes. Residuals are tail appends, so
+// each must stay a small fraction of a worker prompt: non-empty for the
+// tuned families, empty for default, and bounded well under a kilobyte each.
+
+describe("prompt-variance prompt sizes per family row", () => {
+  test("default residual is empty", () => {
+    expect(defaultRow.residual.length).toBe(0);
+  });
+
+  test("every tuned residual stays under 2000 chars / 3000 bytes", () => {
+    for (const row of [museRow, grokRow]) {
+      expect(row.residual.length).toBeGreaterThan(0);
+      expect(row.residual.length).toBeLessThan(2000);
+      expect(Buffer.byteLength(row.residual, "utf8")).toBeLessThan(3000);
+    }
+  });
+
+  test("the grok finish-bias residual is larger than the muse discipline rules", () => {
+    expect(grokRow.residual.length).toBeGreaterThan(museRow.residual.length);
+  });
+});
diff --git a/src/agent/prompt-sizes.test.ts b/src/agent/prompt-sizes.test.ts
index dc3136cf4..4df97e16a 100644
--- a/src/agent/prompt-sizes.test.ts
+++ b/src/agent/prompt-sizes.test.ts
@@ -127,10 +127,10 @@ function budgetMessage(
 describe("director prompt size budget", () => {
   const rows = directorPromptSizeTable();
 
-  test("covers every director in both families", () => {
-    expect(rows.length).toBe(DIRECTOR_IDS.length * 2);
+  test("covers every director in all three variance families", () => {
+    expect(rows.length).toBe(DIRECTOR_IDS.length * 3);
     for (const directorId of DIRECTOR_IDS) {
-      for (const family of ["default", "grok"] as const) {
+      for (const family of ["default", "muse", "grok"] as const) {
         expect(
           rows.some((r) => r.directorId === directorId && r.family === family),
         ).toBe(true);
@@ -177,6 +177,21 @@ describe("director prompt size budget", () => {
     }
   });
 
+  test("muse family appends the shipped tool-discipline rules", () => {
+    for (const directorId of DIRECTOR_IDS) {
+      const base = rows.find(
+        (r) => r.directorId === directorId && r.family === "default",
+      );
+      const muse = rows.find(
+        (r) => r.directorId === directorId && r.family === "muse",
+      );
+      expect(muse?.chars ?? 0).toBeGreaterThan(base?.chars ?? 0);
+      expect(assembleDirectorPrompt(directorId, "muse")).toContain(
+        "Tool discipline:",
+      );
+    }
+  });
+
   test("measurement is deterministic", () => {
     const again = directorPromptSizeTable();
     expect(again.map((r) => r.chars)).toEqual(rows.map((r) => r.chars));

From 71989feb73ced9785889e8349a19a6eb1b06ab26 Mon Sep 17 00:00:00 2001
From: Sawyer <sawyer@dirtroad.dev>
Date: Fri, 18 Sep 2026 15:23:53 -0700
Subject: [PATCH 3/6] feat(prompt-variance): extract family prompt variance as
 a versioned package
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Directors resolve default/muse/grok rows and assemble prompts through
packages/prompt-variance instead of hand-concatenated grokAntiThrash
booleans. Default/grok director output is byte-identical (sha256 match
across all directors); muse rules now also render in prompt-size rows.

Verification: bun test src/agent src/subagent src/prompts.test.ts
packages/prompt-variance/ — 1416 pass, 0 fail. tsc --noEmit reports only
the pre-existing vendor/intx-types semver error, identical on main.
---
 packages/prompt-variance/package.json    | 13 ++++
 packages/prompt-variance/src/assemble.ts | 33 +++++++++
 packages/prompt-variance/src/index.ts    | 16 ++++
 packages/prompt-variance/src/resolve.ts  | 43 +++++++++++
 packages/prompt-variance/src/rows.ts     | 94 ++++++++++++++++++++++++
 src/agent/model-family-policy.ts         |  8 +-
 src/agent/prompt-sizes.test.ts           | 13 ++--
 src/agent/prompt-sizes.ts                | 31 +++++---
 8 files changed, 232 insertions(+), 19 deletions(-)
 create mode 100644 packages/prompt-variance/package.json
 create mode 100644 packages/prompt-variance/src/assemble.ts
 create mode 100644 packages/prompt-variance/src/index.ts
 create mode 100644 packages/prompt-variance/src/resolve.ts
 create mode 100644 packages/prompt-variance/src/rows.ts

diff --git a/packages/prompt-variance/package.json b/packages/prompt-variance/package.json
new file mode 100644
index 000000000..fe6af5a53
--- /dev/null
+++ b/packages/prompt-variance/package.json
@@ -0,0 +1,13 @@
+{
+  "name": "prompt-variance",
+  "version": "1.4.0",
+  "private": true,
+  "type": "module",
+  "license": "SEE LICENSE IN LICENSE.md",
+  "exports": {
+    ".": {
+      "types": "./src/index.ts",
+      "default": "./src/index.ts"
+    }
+  }
+}
diff --git a/packages/prompt-variance/src/assemble.ts b/packages/prompt-variance/src/assemble.ts
new file mode 100644
index 000000000..70b232d7d
--- /dev/null
+++ b/packages/prompt-variance/src/assemble.ts
@@ -0,0 +1,33 @@
+import type { PromptVarianceRow } from "./rows.js";
+
+export interface AssembledPromptVariance {
+  systemPrompt: string;
+  toolNames: readonly string[];
+}
+
+/**
+ * Assemble a prompt from sections plus a family variance row (CL-8269).
+ * The row residual renders last; advertised denies filter the mounted
+ * tool names. Variance is subtractive only: the output names are always
+ * a subset of the input names, in input order.
+ */
+export function assemble(
+  sections: readonly string[],
+  familyRow: PromptVarianceRow,
+  tools: readonly string[],
+): AssembledPromptVariance {
+  if (familyRow.advertisedToolDeny.includes("use_skill")) {
+    throw new Error(
+      `prompt-variance: row "${familyRow.id}" denies use_skill — brief-named skills always load directly`,
+    );
+  }
+  const systemPrompt =
+    familyRow.residual.length > 0
+      ? [...sections, familyRow.residual].join("\n\n")
+      : sections.join("\n\n");
+  const deny = new Set(familyRow.advertisedToolDeny);
+  return {
+    systemPrompt,
+    toolNames: tools.filter((name) => !deny.has(name)),
+  };
+}
diff --git a/packages/prompt-variance/src/index.ts b/packages/prompt-variance/src/index.ts
new file mode 100644
index 000000000..4f39af5b9
--- /dev/null
+++ b/packages/prompt-variance/src/index.ts
@@ -0,0 +1,16 @@
+export { assemble, type AssembledPromptVariance } from "./assemble.js";
+export {
+  applyRowOverride,
+  resolvePromptVariance,
+} from "./resolve.js";
+export {
+  defaultRow,
+  FAMILY_IDS,
+  FAMILY_ROWS,
+  grokRow,
+  museRow,
+  type PromptVarianceFamily,
+  type PromptVarianceRender,
+  type PromptVarianceRow,
+  type PromptVarianceRowOverride,
+} from "./rows.js";
diff --git a/packages/prompt-variance/src/resolve.ts b/packages/prompt-variance/src/resolve.ts
new file mode 100644
index 000000000..03d3111c8
--- /dev/null
+++ b/packages/prompt-variance/src/resolve.ts
@@ -0,0 +1,43 @@
+import {
+  defaultRow,
+  FAMILY_ROWS,
+  type PromptVarianceFamily,
+  type PromptVarianceRow,
+} from "./rows.js";
+
+/**
+ * Apply a row's per-model-id override, if any. Model ids match
+ * case-insensitively against the row's lowercase override keys; an
+ * unknown id resolves to the base row unchanged.
+ */
+export function applyRowOverride(
+  row: PromptVarianceRow,
+  model?: string,
+): PromptVarianceRow {
+  if (model === undefined) return row;
+  const override = row.overrides?.[model.trim().toLowerCase()];
+  if (override === undefined) return row;
+  return {
+    ...row,
+    residual: override.residual ?? row.residual,
+    advertisedToolDeny: override.advertisedToolDeny ?? row.advertisedToolDeny,
+    sectionOmit: override.sectionOmit ?? row.sectionOmit,
+  };
+}
+
+/**
+ * Resolve the variance row for a family, mirroring the leaves-only gate:
+ * the grok finish-bias residual only makes sense on leaf workers, so
+ * orchestrators fall back to the default shape. Muse keeps its residual
+ * on both — its constructors append the same text at the tail either way.
+ */
+export function resolvePromptVariance(input: {
+  family: PromptVarianceFamily;
+  orchestrator?: boolean;
+  model?: string;
+}): PromptVarianceRow {
+  if (input.family === "grok" && input.orchestrator === true) {
+    return applyRowOverride(defaultRow, input.model);
+  }
+  return applyRowOverride(FAMILY_ROWS[input.family], input.model);
+}
diff --git a/packages/prompt-variance/src/rows.ts b/packages/prompt-variance/src/rows.ts
new file mode 100644
index 000000000..dcac93a9d
--- /dev/null
+++ b/packages/prompt-variance/src/rows.ts
@@ -0,0 +1,94 @@
+/**
+ * Versioned model-family prompt variance (CL-8269). One row per tuned
+ * family: a tail residual plus the tool names the family must not
+ * advertise. Directors assemble prompts through `assemble`, never by
+ * hand-concatenating family booleans.
+ *
+ * Only families with shipped eval numbers live here: default/muse/grok.
+ * glm/claude/gpt rows land when CL-8265 / 7772 / 7775 characterize them.
+ */
+
+/** Families with a shipped variance row. */
+export type PromptVarianceFamily = "default" | "muse" | "grok";
+
+/** Residuals render at the tail so they cannot disturb the cached prefix. */
+export type PromptVarianceRender = "tail";
+
+export interface PromptVarianceRowOverride {
+  residual?: string;
+  advertisedToolDeny?: readonly string[];
+  sectionOmit?: readonly string[];
+}
+
+export interface PromptVarianceRow {
+  id: PromptVarianceFamily;
+  render: PromptVarianceRender;
+  /** Tail text appended to the assembled prompt. Empty when the family needs none. */
+  residual: string;
+  /**
+   * Tool names to drop from the advertised set (CL-7668). Subtractive
+   * only — a row can never grant a tool `assemble` was not given.
+   * `use_skill` is never denied: brief-named skills load directly.
+   */
+  advertisedToolDeny: readonly string[];
+  /** Assembly section ids to omit for this family. Empty until a family needs one. */
+  sectionOmit: readonly string[];
+  /** Optional per-model-id refinements, keyed by lowercase model id. */
+  overrides?: Record<string, PromptVarianceRowOverride>;
+}
+
+export const FAMILY_IDS: readonly PromptVarianceFamily[] = [
+  "default",
+  "muse",
+  "grok",
+];
+
+export const defaultRow: PromptVarianceRow = {
+  id: "default",
+  render: "tail",
+  residual: "",
+  advertisedToolDeny: [],
+  sectionOmit: [],
+};
+
+// Muse Spark does not reliably stop a tool loop at medium reasoning effort:
+// the same run with these three rules appended finished in 3 turns on 4.3x
+// fewer input tokens (CL-7869). Byte-identical to the shipped text so the
+// size table and the constructor append cannot drift apart.
+export const museRow: PromptVarianceRow = {
+  id: "muse",
+  render: "tail",
+  residual:
+    "Tool discipline:\n" +
+    "- Batch independent tool calls into a single turn.\n" +
+    "- Never re-read a file you have already read this session.\n" +
+    "- Do not narrate; act.",
+  advertisedToolDeny: [],
+  sectionOmit: [],
+};
+
+// Tiny residual for Grok/xAI workers: mining showed higher tools-only thrash
+// than Codex on the same harness. Shared thrash harness + spawn contracts do
+// the structural work; this is only a finish-bias nudge, not a full rewrite.
+// Built from the current grok finish-bias text on origin/main; the CL-8297
+// residual folds in here at merge time.
+export const grokRow: PromptVarianceRow = {
+  id: "grok",
+  render: "tail",
+  residual: [
+    "Finish bias (xAI / Grok worker):",
+    "- Once you can answer the dispatch brief, prefer the structured report over another speculative tool call.",
+    "- If the next call would only re-open paths you already read, write the report instead.",
+    "- When the dispatch brief's done-definition is met, write the report envelope instead of making one more search or micro-edit.",
+    "- Route file and web work through the dedicated tools, never run_shell — mining showed grok reaching for shell first when a typed tool already covered the job.",
+  ].join("\n"),
+  // Leaf value; resolvePromptVariance clears it for orchestrators.
+  advertisedToolDeny: ["skill_search"],
+  sectionOmit: [],
+};
+
+export const FAMILY_ROWS: Record<PromptVarianceFamily, PromptVarianceRow> = {
+  default: defaultRow,
+  muse: museRow,
+  grok: grokRow,
+};
diff --git a/src/agent/model-family-policy.ts b/src/agent/model-family-policy.ts
index 870f7ec45..721887b33 100644
--- a/src/agent/model-family-policy.ts
+++ b/src/agent/model-family-policy.ts
@@ -2,6 +2,7 @@ import {
   detectModelFamily,
   type ModelFamily,
 } from "../subagent/provider-family.js";
+import { museRow } from "../../packages/prompt-variance/src/index.js";
 
 /**
  * Per-model-family tuning for the shared directors (main chat director and
@@ -120,11 +121,8 @@ const KIMI_POLICY: Omit<ModelFamilyPolicy, "family"> = { ...DEFAULT_POLICY };
 // three rules appended finished in 3 turns on 4.3x fewer input tokens. At
 // minimal effort it terminates either way, so the rules earn their keep exactly
 // at the rungs where each wasted turn is most expensive. See CL-7869.
-const MUSE_TOOL_DISCIPLINE_RULES =
-  "Tool discipline:\n" +
-  "- Batch independent tool calls into a single turn.\n" +
-  "- Never re-read a file you have already read this session.\n" +
-  "- Do not narrate; act.";
+// Single-sourced from the versioned prompt-variance package (CL-8269).
+const MUSE_TOOL_DISCIPLINE_RULES = museRow.residual;
 
 const MUSE_POLICY: Omit<ModelFamilyPolicy, "family"> = {
   ...DEFAULT_POLICY,
diff --git a/src/agent/prompt-sizes.test.ts b/src/agent/prompt-sizes.test.ts
index 4df97e16a..5c2985f1d 100644
--- a/src/agent/prompt-sizes.test.ts
+++ b/src/agent/prompt-sizes.test.ts
@@ -27,7 +27,7 @@ import { resolveExecDirectorOverlay } from "../exec/runner.js";
 /**
  * Prompt size budget (CL-7664). Numeric asserts only — copy edits must not
  * fail this test. Baselines are a checked-in snapshot of the max measured
- * sizes across both families from the canonical fixture in
+ * sizes across all three families from the canonical fixture in
  * src/agent/prompt-sizes.ts; budgets add a +2000 char / +3000 byte allowance
  * (ceiling to 100) in code below. Bytes get the larger headroom because
  * multibyte copy can shift them faster. Adding a director is a type error
@@ -200,7 +200,7 @@ describe("director prompt size budget", () => {
 
   test("tool names match the production mount: no dupes, no phantoms", () => {
     for (const directorId of DIRECTOR_IDS) {
-      for (const family of ["default", "grok"] as const) {
+      for (const family of ["default", "muse", "grok"] as const) {
         const names = canonicalToolNamesForDirector(
           DIRECTOR_REGISTRY[directorId],
           family,
@@ -208,7 +208,7 @@ describe("director prompt size budget", () => {
         expect(new Set(names).size, `${directorId} [${family}]`).toBe(
           names.length,
         );
-        // Neither fixture family is Codex, so the Codex proxies
+        // No fixture family is Codex, so the Codex proxies
         // (createCodexToolProxies returns [] when !isCodex) must be absent,
         // as must list_dir, which no subagent mount installs.
         for (const phantom of [
@@ -234,17 +234,20 @@ describe("director prompt size budget", () => {
   test("formatPromptSizeTable renders one row per director", () => {
     const table = formatPromptSizeTable(rows);
     expect(table).toContain(
-      "| director | default chars (bytes) | grok chars (bytes) |",
+      "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) |",
     );
     for (const directorId of DIRECTOR_IDS) {
       const base = rows.find(
         (r) => r.directorId === directorId && r.family === "default",
       );
+      const muse = rows.find(
+        (r) => r.directorId === directorId && r.family === "muse",
+      );
       const grok = rows.find(
         (r) => r.directorId === directorId && r.family === "grok",
       );
       expect(table).toContain(
-        `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
+        `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
       );
     }
   });
diff --git a/src/agent/prompt-sizes.ts b/src/agent/prompt-sizes.ts
index 8a2944656..b6ce09f46 100644
--- a/src/agent/prompt-sizes.ts
+++ b/src/agent/prompt-sizes.ts
@@ -38,8 +38,9 @@ import { webSearchDefinition } from "../tools/web-search.js";
  * appendix, with the Grok finish-bias note gated by
  * shouldApplyGrokAntiThrash (leaves on Grok-family providers only) and the
  * family promptResidual (CL-8297 tool budget for grok leaves, XML
- * task_guidance block for claude leaves) resolved from the model family
- * policy.
+ * task_guidance block for claude leaves, narrate-before-tools nudge for
+ * gpt leaves) resolved from the model family policy. Residual texts are
+ * single-sourced from the versioned prompt-variance package (CL-8269).
  *
  * The env and provider inputs are pinned here so sizes never drift with the
  * machine, date, or checkout — only real prompt changes move the numbers.
@@ -60,13 +61,17 @@ const GROK_PROVIDER = { providerName: "xai/default", model: "grok-4.6" };
 // Default-family probe: an unrecognized provider stays on the default
 // family no matter how many family rows land (claude/gpt already ship),
 // so the default column carries no residual.
+const MUSE_PROVIDER = {
+  providerName: "opencode-go",
+  model: "muse-spark-1.3-contributor",
+};
 const DEFAULT_PROVIDER = {
   providerName: "unknown-provider",
   model: "unknown-model",
 };
 
-/** Families in the size table: default assembly vs Grok (+finish-bias note). */
-export type PromptSizeFamily = "default" | "grok";
+/** Families in the size table: default, muse (+tool-discipline rules), grok (+finish-bias note). */
+export type PromptSizeFamily = "default" | "muse" | "grok";
 
 /**
  * Pinned AGENTS.md body for prefix measurement. Production reads the live
@@ -162,7 +167,12 @@ export function assembleDirectorPrompt(
 ): string {
   const pkg = DIRECTOR_REGISTRY[directorId];
   const orchestrator = pkg.spawn.maySpawn;
-  const provider = family === "grok" ? GROK_PROVIDER : DEFAULT_PROVIDER;
+  const provider =
+    family === "grok"
+      ? GROK_PROVIDER
+      : family === "muse"
+        ? MUSE_PROVIDER
+        : DEFAULT_PROVIDER;
   const policy = resolveModelFamilyPolicy({ ...provider, orchestrator });
   return buildSubAgentSystemPrompt(
     [formatDirectorSystemPrompt(pkg)],
@@ -234,7 +244,7 @@ export function measureSkywalkerPrefix(): SkywalkerPrefixSize {
 export function directorPromptSizeTable(): DirectorPromptSize[] {
   const rows: DirectorPromptSize[] = [];
   for (const directorId of DIRECTOR_IDS) {
-    for (const family of ["default", "grok"] as const) {
+    for (const family of ["default", "muse", "grok"] as const) {
       rows.push(measureDirectorPrompt(directorId, family));
     }
   }
@@ -244,18 +254,21 @@ export function directorPromptSizeTable(): DirectorPromptSize[] {
 /** Render the size table as markdown (for PR bodies and budget updates). */
 export function formatPromptSizeTable(rows: DirectorPromptSize[]): string {
   const lines = [
-    "| director | default chars (bytes) | grok chars (bytes) |",
-    "| --- | --- | --- |",
+    "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) |",
+    "| --- | --- | --- | --- |",
   ];
   for (const directorId of DIRECTOR_IDS) {
     const base = rows.find(
       (r) => r.directorId === directorId && r.family === "default",
     );
+    const muse = rows.find(
+      (r) => r.directorId === directorId && r.family === "muse",
+    );
     const grok = rows.find(
       (r) => r.directorId === directorId && r.family === "grok",
     );
     lines.push(
-      `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
+      `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
     );
   }
   return lines.join("\n");

From 60d81ef7550d0b20b2bddb28764ce0feb145249f Mon Sep 17 00:00:00 2001
From: Sawyer <sawyer@dirtroad.dev>
Date: Fri, 18 Sep 2026 15:28:26 -0700
Subject: [PATCH 4/6] chore(prompt-variance): sync bun.lock for new workspace
 package and apply oxfmt

---
 bun.lock                                  | 6 ++++++
 packages/prompt-variance/src/index.ts     | 5 +----
 packages/prompt-variance/src/rows.test.ts | 6 +-----
 3 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/bun.lock b/bun.lock
index bd1848922..60419df85 100644
--- a/bun.lock
+++ b/bun.lock
@@ -59,6 +59,10 @@
       "name": "@corbits/provider-opencode-go",
       "version": "0.1.0",
     },
+    "packages/prompt-variance": {
+      "name": "prompt-variance",
+      "version": "1.4.0",
+    },
     "packages/zen": {
       "name": "@corbits/provider-zen",
       "version": "0.1.0",
@@ -750,6 +754,8 @@
 
     "process": ["process@0.11.10", "", {}, "sha512-cdGef/drWFoydD1JsMzuFf8100nZl+GT+yacc2bEced5f9Rjk4z+WtFUTBu9PhOi9j/jfmBPu0mMEY4wIdAF8A=="],
 
+    "prompt-variance": ["prompt-variance@workspace:packages/prompt-variance"],
+
     "proxy-addr": ["proxy-addr@2.0.7", "", { "dependencies": { "forwarded": "0.2.0", "ipaddr.js": "1.9.1" } }, "sha512-llQsMLSUDUPT44jdrU/O37qlnifitDP+ZwrmmZcoSKyLKvtZxpyV0n2/bD/N4tBAAZ/gJEdZU7KMraoK1+XYAg=="],
 
     "qs": ["qs@6.16.0", "", { "dependencies": { "es-define-property": "^1.0.1", "side-channel": "^1.1.1" } }, "sha512-h6fhOIaRrID2CbEY2fqs+7t+UXZo+MLAnU5gRIq85uFtdiUPCdsApMlHhXogKVM4HM2DVbIjGNTTYH2OcmP1vA=="],
diff --git a/packages/prompt-variance/src/index.ts b/packages/prompt-variance/src/index.ts
index 4f39af5b9..ffc2f3a02 100644
--- a/packages/prompt-variance/src/index.ts
+++ b/packages/prompt-variance/src/index.ts
@@ -1,8 +1,5 @@
 export { assemble, type AssembledPromptVariance } from "./assemble.js";
-export {
-  applyRowOverride,
-  resolvePromptVariance,
-} from "./resolve.js";
+export { applyRowOverride, resolvePromptVariance } from "./resolve.js";
 export {
   defaultRow,
   FAMILY_IDS,
diff --git a/packages/prompt-variance/src/rows.test.ts b/packages/prompt-variance/src/rows.test.ts
index 18441687d..d7ce985b4 100644
--- a/packages/prompt-variance/src/rows.test.ts
+++ b/packages/prompt-variance/src/rows.test.ts
@@ -31,11 +31,7 @@ describe("prompt-variance family rows", () => {
   });
 
   test("row ids match their family", () => {
-    const ids: PromptVarianceFamily[] = [
-      defaultRow.id,
-      museRow.id,
-      grokRow.id,
-    ];
+    const ids: PromptVarianceFamily[] = [defaultRow.id, museRow.id, grokRow.id];
     expect(ids).toEqual(["default", "muse", "grok"]);
   });
 

From 88c9fb04c71172eba36d49a91d4758fbc745c615 Mon Sep 17 00:00:00 2001
From: Sawyer <sawyer@dirtroad.dev>
Date: Fri, 18 Sep 2026 15:35:18 -0700
Subject: [PATCH 5/6] chore(prompt-variance): allowlist versioned barrel
 surface for dead-export guard

---
 scripts/dead-export-allowlist.txt | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/scripts/dead-export-allowlist.txt b/scripts/dead-export-allowlist.txt
index 91221f98a..3f3ddf02f 100644
--- a/scripts/dead-export-allowlist.txt
+++ b/scripts/dead-export-allowlist.txt
@@ -49,3 +49,21 @@ src/agent/directors/identity.ts: satisfies
 src/agent/directors/identity.ts: Record
 packages/opencode-go/src/models.ts: satisfies
 packages/opencode-go/src/models.ts: readonly
+
+# Versioned prompt-variance package public surface (CL-8269 lane,
+# agent-owned). Family rows data, the assemble shape, and per-model-id
+# overrides re-exported through the package entry point for the 0.3.31+
+# agent packages; no in-repo barrel consumer exists yet (src/ consumes
+# assemble, resolvePromptVariance, grokRow, museRow, and PromptVarianceRow
+# only, and the package tests import the backing modules directly). Scoped
+# to the exact ts-prune flags so a new dead export in this barrel fails the
+# guard instead of hiding under a whole-file exemption. Remove each entry
+# with the export it names if the public surface shrinks.
+packages/prompt-variance/src/index.ts: AssembledPromptVariance
+packages/prompt-variance/src/index.ts: applyRowOverride
+packages/prompt-variance/src/index.ts: defaultRow
+packages/prompt-variance/src/index.ts: FAMILY_IDS
+packages/prompt-variance/src/index.ts: FAMILY_ROWS
+packages/prompt-variance/src/index.ts: PromptVarianceFamily
+packages/prompt-variance/src/index.ts: PromptVarianceRender
+packages/prompt-variance/src/index.ts: PromptVarianceRowOverride

From 035835a69180fc37fda22d706a2b5f1f9cb76b37 Mon Sep 17 00:00:00 2001
From: Sawyer Cutler <sawyer@dirtroad.dev>
Date: Sun, 20 Sep 2026 20:42:26 -0700
Subject: [PATCH 6/6] feat(prompt-variance): residuals-only rows as single
 source for five families

Deny decision (default residuals-only): the package owns residual TEXT;
tool denial stays live in ModelFamilyPolicy.advertisedToolDeny, applied
at mount time by run.ts. Drops advertisedToolDeny, sectionOmit,
per-model overrides, and the AssembledPromptVariance shape.

Fold tool-budget (CL-8297) + ceremony (CL-8296) + claude (CL-8309) + gpt
(CL-8310) rows into rows.ts; policy residual names alias the rows.
Package renamed to @corbits/prompt-variance at 0.1.0; bun.lock synced.
Size table gains claude/gpt columns; fixture mirrors the director
tail-append for muse discipline rules.
---
 bun.lock                                      |   8 +-
 packages/prompt-variance/package.json         |   4 +-
 packages/prompt-variance/src/assemble.test.ts |  50 +++-----
 packages/prompt-variance/src/assemble.ts      |  33 ++----
 packages/prompt-variance/src/index.ts         |   9 +-
 packages/prompt-variance/src/resolve.test.ts  |  55 ++++-----
 packages/prompt-variance/src/resolve.ts       |  39 ++----
 packages/prompt-variance/src/rows.test.ts     |  99 +++++++++++-----
 packages/prompt-variance/src/rows.ts          | 111 +++++++++++-------
 packages/prompt-variance/src/sizes.test.ts    |   8 +-
 scripts/dead-export-allowlist.txt             |  16 ++-
 src/agent/model-family-policy.ts              |  53 ++++-----
 src/agent/prompt-sizes.test.ts                |  62 +++++++++-
 src/agent/prompt-sizes.ts                     |  57 +++++++--
 14 files changed, 342 insertions(+), 262 deletions(-)

diff --git a/bun.lock b/bun.lock
index 60419df85..5f08b55f8 100644
--- a/bun.lock
+++ b/bun.lock
@@ -60,8 +60,8 @@
       "version": "0.1.0",
     },
     "packages/prompt-variance": {
-      "name": "prompt-variance",
-      "version": "1.4.0",
+      "name": "@corbits/prompt-variance",
+      "version": "0.1.0",
     },
     "packages/zen": {
       "name": "@corbits/provider-zen",
@@ -228,6 +228,8 @@
 
     "@corbits/openai-responses": ["@corbits/openai-responses@github:corbitsdev/corbits-openai-responses#1d20dbb", { "dependencies": { "arktype": "2.2.3" }, "peerDependencies": { "@intx/inference": ">=0.3.0", "@intx/types": ">=0.3.0" } }, "corbitsdev-corbits-openai-responses-1d20dbb", "sha512-lSJz1KkKD8mEUYiQ57dJprD9JavdJGt3OBRa42tvXLsikt1fJ6qX2mGCwaEXzkfZNc2XY5o7ZpxK5swSJX+LnA=="],
 
+    "@corbits/prompt-variance": ["@corbits/prompt-variance@workspace:packages/prompt-variance"],
+
     "@corbits/provider-opencode-go": ["@corbits/provider-opencode-go@workspace:packages/opencode-go"],
 
     "@corbits/provider-zen": ["@corbits/provider-zen@workspace:packages/zen"],
@@ -754,8 +756,6 @@
 
     "process": ["process@0.11.10", "", {}, "sha512-cdGef/drWFoydD1JsMzuFf8100nZl+GT+yacc2bEced5f9Rjk4z+WtFUTBu9PhOi9j/jfmBPu0mMEY4wIdAF8A=="],
 
-    "prompt-variance": ["prompt-variance@workspace:packages/prompt-variance"],
-
     "proxy-addr": ["proxy-addr@2.0.7", "", { "dependencies": { "forwarded": "0.2.0", "ipaddr.js": "1.9.1" } }, "sha512-llQsMLSUDUPT44jdrU/O37qlnifitDP+ZwrmmZcoSKyLKvtZxpyV0n2/bD/N4tBAAZ/gJEdZU7KMraoK1+XYAg=="],
 
     "qs": ["qs@6.16.0", "", { "dependencies": { "es-define-property": "^1.0.1", "side-channel": "^1.1.1" } }, "sha512-h6fhOIaRrID2CbEY2fqs+7t+UXZo+MLAnU5gRIq85uFtdiUPCdsApMlHhXogKVM4HM2DVbIjGNTTYH2OcmP1vA=="],
diff --git a/packages/prompt-variance/package.json b/packages/prompt-variance/package.json
index fe6af5a53..714ecc30f 100644
--- a/packages/prompt-variance/package.json
+++ b/packages/prompt-variance/package.json
@@ -1,6 +1,6 @@
 {
-  "name": "prompt-variance",
-  "version": "1.4.0",
+  "name": "@corbits/prompt-variance",
+  "version": "0.1.0",
   "private": true,
   "type": "module",
   "license": "SEE LICENSE IN LICENSE.md",
diff --git a/packages/prompt-variance/src/assemble.test.ts b/packages/prompt-variance/src/assemble.test.ts
index a7e2b3ec8..a0c0aeeca 100644
--- a/packages/prompt-variance/src/assemble.test.ts
+++ b/packages/prompt-variance/src/assemble.test.ts
@@ -1,52 +1,36 @@
 import { describe, expect, test } from "bun:test";
 import { assemble } from "./assemble.js";
-import { defaultRow, grokRow, museRow } from "./rows.js";
-
-// CL-8269 RED: assemble does not exist yet — every test below fails until
-// the GREEN lands packages/prompt-variance.
+import { claudeRow, defaultRow, gptRow, grokRow, museRow } from "./rows.js";
 
 const SECTIONS = ["contract", "Tools (names only): read_file", "context"];
-const TOOLS = ["read_file", "skill_search", "use_skill", "run_shell"];
 
 describe("assemble", () => {
   test("joins sections with the residual as the last section", () => {
-    const out = assemble(SECTIONS, grokRow, TOOLS);
-    expect(out.systemPrompt).toContain("contract");
-    expect(out.systemPrompt.trimEnd().endsWith(grokRow.residual)).toBe(true);
+    const out = assemble(SECTIONS, grokRow);
+    expect(out).toContain("contract");
+    expect(out.trimEnd().endsWith(grokRow.residual)).toBe(true);
   });
 
   test("a row without residual leaves the sections unchanged", () => {
-    const out = assemble(SECTIONS, defaultRow, TOOLS);
-    expect(out.systemPrompt).toBe(SECTIONS.join("\n\n"));
+    expect(assemble(SECTIONS, defaultRow)).toBe(SECTIONS.join("\n\n"));
   });
 
   test("muse residual closes the prompt like the shipped tail append", () => {
-    const out = assemble(SECTIONS, museRow, TOOLS);
-    expect(out.systemPrompt.trimEnd().endsWith(museRow.residual)).toBe(true);
-  });
-
-  test("filters advertisedToolDeny from the tool names, preserving order", () => {
-    const out = assemble(SECTIONS, grokRow, TOOLS);
-    expect([...out.toolNames]).toEqual(["read_file", "use_skill", "run_shell"]);
-  });
-
-  test("variance cannot grant tools: output is always a subset of the input", () => {
-    for (const row of [defaultRow, museRow, grokRow]) {
-      const out = assemble(SECTIONS, row, TOOLS);
-      for (const name of out.toolNames) {
-        expect(TOOLS).toContain(name);
-      }
-      expect(out.toolNames.length).toBeLessThanOrEqual(TOOLS.length);
-    }
+    const out = assemble(SECTIONS, museRow);
+    expect(out.trimEnd().endsWith(museRow.residual)).toBe(true);
   });
 
-  test("variance cannot grant tools: unknown mounted names pass through untouched", () => {
-    const out = assemble(SECTIONS, defaultRow, ["alpha", "beta"]);
-    expect([...out.toolNames]).toEqual(["alpha", "beta"]);
+  test("claude and gpt residuals close the prompt the same way", () => {
+    expect(
+      assemble(SECTIONS, claudeRow).trimEnd().endsWith(claudeRow.residual),
+    ).toBe(true);
+    expect(assemble(SECTIONS, gptRow).trimEnd().endsWith(gptRow.residual)).toBe(
+      true,
+    );
   });
 
-  test("refuses a row that denies use_skill", () => {
-    const bad = { ...grokRow, advertisedToolDeny: ["use_skill"] };
-    expect(() => assemble(SECTIONS, bad, TOOLS)).toThrow(/use_skill/);
+  test("tool mounting is out of scope: assemble returns a string, not tool names", () => {
+    const out = assemble(SECTIONS, grokRow);
+    expect(typeof out).toBe("string");
   });
 });
diff --git a/packages/prompt-variance/src/assemble.ts b/packages/prompt-variance/src/assemble.ts
index 70b232d7d..fd56fc1df 100644
--- a/packages/prompt-variance/src/assemble.ts
+++ b/packages/prompt-variance/src/assemble.ts
@@ -1,33 +1,16 @@
 import type { PromptVarianceRow } from "./rows.js";
 
-export interface AssembledPromptVariance {
-  systemPrompt: string;
-  toolNames: readonly string[];
-}
-
 /**
- * Assemble a prompt from sections plus a family variance row (CL-8269).
- * The row residual renders last; advertised denies filter the mounted
- * tool names. Variance is subtractive only: the output names are always
- * a subset of the input names, in input order.
+ * Append a family variance row's residual at the tail of the assembled
+ * sections (CL-8269). An empty residual leaves the sections unchanged.
+ * Residuals render last so they cannot disturb the cached prompt prefix.
+ * Tool mounting is out of scope: run.ts applies
+ * ModelFamilyPolicy.advertisedToolDeny at mount time.
  */
 export function assemble(
   sections: readonly string[],
   familyRow: PromptVarianceRow,
-  tools: readonly string[],
-): AssembledPromptVariance {
-  if (familyRow.advertisedToolDeny.includes("use_skill")) {
-    throw new Error(
-      `prompt-variance: row "${familyRow.id}" denies use_skill — brief-named skills always load directly`,
-    );
-  }
-  const systemPrompt =
-    familyRow.residual.length > 0
-      ? [...sections, familyRow.residual].join("\n\n")
-      : sections.join("\n\n");
-  const deny = new Set(familyRow.advertisedToolDeny);
-  return {
-    systemPrompt,
-    toolNames: tools.filter((name) => !deny.has(name)),
-  };
+): string {
+  if (familyRow.residual.length === 0) return sections.join("\n\n");
+  return [...sections, familyRow.residual].join("\n\n");
 }
diff --git a/packages/prompt-variance/src/index.ts b/packages/prompt-variance/src/index.ts
index ffc2f3a02..43b010f03 100644
--- a/packages/prompt-variance/src/index.ts
+++ b/packages/prompt-variance/src/index.ts
@@ -1,13 +1,14 @@
-export { assemble, type AssembledPromptVariance } from "./assemble.js";
-export { applyRowOverride, resolvePromptVariance } from "./resolve.js";
+export { assemble } from "./assemble.js";
+export { resolvePromptVariance } from "./resolve.js";
 export {
+  claudeRow,
   defaultRow,
   FAMILY_IDS,
   FAMILY_ROWS,
+  gptRow,
   grokRow,
+  grokToolBudgetResidual,
   museRow,
   type PromptVarianceFamily,
-  type PromptVarianceRender,
   type PromptVarianceRow,
-  type PromptVarianceRowOverride,
 } from "./rows.js";
diff --git a/packages/prompt-variance/src/resolve.test.ts b/packages/prompt-variance/src/resolve.test.ts
index f98786ed5..8e071683c 100644
--- a/packages/prompt-variance/src/resolve.test.ts
+++ b/packages/prompt-variance/src/resolve.test.ts
@@ -1,27 +1,37 @@
 import { describe, expect, test } from "bun:test";
-import { applyRowOverride, resolvePromptVariance } from "./resolve.js";
+import { resolvePromptVariance } from "./resolve.js";
 import { grokRow } from "./rows.js";
 
-// CL-8269 RED: the resolver does not exist yet — every test below fails
-// until the GREEN lands packages/prompt-variance.
-
 describe("resolvePromptVariance", () => {
   test("resolves each family to its row", () => {
     expect(resolvePromptVariance({ family: "default" }).id).toBe("default");
     expect(resolvePromptVariance({ family: "muse" }).id).toBe("muse");
     expect(resolvePromptVariance({ family: "grok" }).id).toBe("grok");
+    expect(resolvePromptVariance({ family: "claude" }).id).toBe("claude");
+    expect(resolvePromptVariance({ family: "gpt" }).id).toBe("gpt");
   });
 
-  test("grok keeps its finish-bias residual and skill_search deny on leaves", () => {
+  test("grok keeps its finish-bias residual on leaves", () => {
     const row = resolvePromptVariance({ family: "grok", orchestrator: false });
     expect(row.residual).toBe(grokRow.residual);
-    expect(row.advertisedToolDeny).toContain("skill_search");
   });
 
-  test("grok on orchestrators falls back to the default shape", () => {
+  test("grok on orchestrators falls back to the default row", () => {
     const row = resolvePromptVariance({ family: "grok", orchestrator: true });
+    expect(row.id).toBe("default");
     expect(row.residual).toBe("");
-    expect([...row.advertisedToolDeny]).toEqual([]);
+  });
+
+  test("claude keeps its task_guidance block on leaves, not orchestrators", () => {
+    expect(
+      resolvePromptVariance({ family: "claude", orchestrator: false }).residual,
+    ).toContain("<task_guidance>");
+    const orch = resolvePromptVariance({
+      family: "claude",
+      orchestrator: true,
+    });
+    expect(orch.id).toBe("default");
+    expect(orch.residual).toBe("");
   });
 
   test("muse keeps its residual on leaves and orchestrators alike", () => {
@@ -33,27 +43,12 @@ describe("resolvePromptVariance", () => {
     ).not.toBe("");
   });
 
-  test("applies per-model-id overrides from the row", () => {
-    const row = applyRowOverride(
-      {
-        ...grokRow,
-        overrides: {
-          "grok-4-special": { advertisedToolDeny: [] },
-        },
-      },
-      "GROK-4-SPECIAL",
-    );
-    expect([...row.advertisedToolDeny]).toEqual([]);
-    expect(row.residual).toBe(grokRow.residual);
-  });
-
-  test("an unknown model id resolves to the base row", () => {
-    const row = resolvePromptVariance({
-      family: "grok",
-      orchestrator: false,
-      model: "grok-9-unknown",
-    });
-    expect(row.residual).toBe(grokRow.residual);
-    expect([...row.advertisedToolDeny]).toEqual(["skill_search"]);
+  test("gpt keeps its narrate-before-tools nudge on leaves and orchestrators alike", () => {
+    expect(
+      resolvePromptVariance({ family: "gpt", orchestrator: false }).residual,
+    ).toContain("Narrate before tools");
+    expect(
+      resolvePromptVariance({ family: "gpt", orchestrator: true }).residual,
+    ).toContain("Narrate before tools");
   });
 });
diff --git a/packages/prompt-variance/src/resolve.ts b/packages/prompt-variance/src/resolve.ts
index 03d3111c8..d3e7d100d 100644
--- a/packages/prompt-variance/src/resolve.ts
+++ b/packages/prompt-variance/src/resolve.ts
@@ -6,38 +6,21 @@ import {
 } from "./rows.js";
 
 /**
- * Apply a row's per-model-id override, if any. Model ids match
- * case-insensitively against the row's lowercase override keys; an
- * unknown id resolves to the base row unchanged.
- */
-export function applyRowOverride(
-  row: PromptVarianceRow,
-  model?: string,
-): PromptVarianceRow {
-  if (model === undefined) return row;
-  const override = row.overrides?.[model.trim().toLowerCase()];
-  if (override === undefined) return row;
-  return {
-    ...row,
-    residual: override.residual ?? row.residual,
-    advertisedToolDeny: override.advertisedToolDeny ?? row.advertisedToolDeny,
-    sectionOmit: override.sectionOmit ?? row.sectionOmit,
-  };
-}
-
-/**
- * Resolve the variance row for a family, mirroring the leaves-only gate:
- * the grok finish-bias residual only makes sense on leaf workers, so
- * orchestrators fall back to the default shape. Muse keeps its residual
- * on both — its constructors append the same text at the tail either way.
+ * Resolve the variance row for a family, mirroring the leaves-only gates in
+ * the model family policy: the grok finish-bias residual and the claude
+ * task_guidance block only make sense on leaf workers, so orchestrators
+ * fall back to the default row. Muse keeps its residual on both — the
+ * constructors append the same text at the tail either way — and gpt keeps
+ * its narrate-before-tools nudge on primaries and leaves alike.
  */
 export function resolvePromptVariance(input: {
   family: PromptVarianceFamily;
   orchestrator?: boolean;
-  model?: string;
 }): PromptVarianceRow {
-  if (input.family === "grok" && input.orchestrator === true) {
-    return applyRowOverride(defaultRow, input.model);
+  if (input.orchestrator === true) {
+    if (input.family === "grok" || input.family === "claude") {
+      return defaultRow;
+    }
   }
-  return applyRowOverride(FAMILY_ROWS[input.family], input.model);
+  return FAMILY_ROWS[input.family];
 }
diff --git a/packages/prompt-variance/src/rows.test.ts b/packages/prompt-variance/src/rows.test.ts
index d7ce985b4..5376b4e2d 100644
--- a/packages/prompt-variance/src/rows.test.ts
+++ b/packages/prompt-variance/src/rows.test.ts
@@ -1,38 +1,65 @@
 import { describe, expect, test } from "bun:test";
 import {
+  claudeRow,
   defaultRow,
   FAMILY_IDS,
+  FAMILY_ROWS,
+  gptRow,
   grokRow,
+  grokToolBudgetResidual,
   museRow,
   type PromptVarianceFamily,
 } from "./rows.js";
 
-// CL-8269 RED: the versioned prompt-variance package does not exist yet, so
-// every test below fails at import time until the GREEN lands it.
+const CEREMONY_LINES = [
+  "Never run git add, git commit, git stash, or any other state-changing git command unless the user asks.",
+  "Do not narrate a plan before acting on a small task; act, then report.",
+  "Verify with the test command once at the end, not after every edit.",
+];
 
 describe("prompt-variance family rows", () => {
-  test("ships exactly the default/muse/grok families", () => {
-    expect([...FAMILY_IDS]).toEqual(["default", "muse", "grok"]);
+  test("ships exactly the default/muse/grok/claude/gpt families", () => {
+    expect([...FAMILY_IDS]).toEqual([
+      "default",
+      "muse",
+      "grok",
+      "claude",
+      "gpt",
+    ]);
   });
 
-  test("has no glm/claude/gpt rows until their evals land (CL-8265/7772/7775)", () => {
+  test("has no glm row until its eval lands (CL-8265)", () => {
     for (const id of FAMILY_IDS) {
-      expect(["glm", "claude", "gpt"]).not.toContain(id);
+      expect(id).not.toBe("glm");
     }
   });
 
-  test("every row carries the render/residual/deny/omit shape", () => {
-    for (const row of [defaultRow, museRow, grokRow]) {
-      expect(row.render).toBe("tail");
+  test("every row carries the id/residual shape — residuals-only, no deny fields", () => {
+    for (const row of [defaultRow, museRow, grokRow, claudeRow, gptRow]) {
+      expect(typeof row.id).toBe("string");
       expect(typeof row.residual).toBe("string");
-      expect(Array.isArray([...row.advertisedToolDeny])).toBe(true);
-      expect(Array.isArray([...row.sectionOmit])).toBe(true);
+      expect("advertisedToolDeny" in row).toBe(false);
+      expect("sectionOmit" in row).toBe(false);
+      expect("overrides" in row).toBe(false);
     }
   });
 
-  test("row ids match their family", () => {
-    const ids: PromptVarianceFamily[] = [defaultRow.id, museRow.id, grokRow.id];
-    expect(ids).toEqual(["default", "muse", "grok"]);
+  test("row ids match their family and FAMILY_ROWS covers all five", () => {
+    const ids: PromptVarianceFamily[] = [
+      defaultRow.id,
+      museRow.id,
+      grokRow.id,
+      claudeRow.id,
+      gptRow.id,
+    ];
+    expect(ids).toEqual(["default", "muse", "grok", "claude", "gpt"]);
+    expect(Object.keys(FAMILY_ROWS).sort()).toEqual([
+      "claude",
+      "default",
+      "gpt",
+      "grok",
+      "muse",
+    ]);
   });
 
   test("muse row is the shipped CL-7869 tool-discipline text", () => {
@@ -42,7 +69,7 @@ describe("prompt-variance family rows", () => {
     expect(museRow.residual).toContain("Do not narrate; act.");
   });
 
-  test("grok row is the existing finish-bias residual", () => {
+  test("grok row is the single finish-bias + ceremony block (CL-8296)", () => {
     expect(grokRow.residual).toContain("Finish bias (xAI / Grok worker):");
     expect(grokRow.residual).toContain("prefer the structured report");
     expect(grokRow.residual).toContain("re-open paths you already read");
@@ -50,25 +77,39 @@ describe("prompt-variance family rows", () => {
     expect(grokRow.residual).toContain("never run_shell");
   });
 
-  test("default row carries no residual", () => {
-    expect(defaultRow.residual).toBe("");
+  test("each ceremony line appears exactly once in the grok row (P2 invariant)", () => {
+    for (const line of CEREMONY_LINES) {
+      const occurrences = grokRow.residual.split(line).length - 1;
+      expect(occurrences).toBe(1);
+    }
   });
 
-  test("grok leaves deny skill_search only; other rows deny nothing", () => {
-    expect([...grokRow.advertisedToolDeny]).toEqual(["skill_search"]);
-    expect([...defaultRow.advertisedToolDeny]).toEqual([]);
-    expect([...museRow.advertisedToolDeny]).toEqual([]);
+  test("grok tool-budget residual is ceremony-free and budget text appears once (P1 invariant)", () => {
+    expect(grokToolBudgetResidual).toContain("Tool budget:");
+    for (const line of CEREMONY_LINES) {
+      expect(grokToolBudgetResidual).not.toContain(line);
+    }
+    const budgetOccurrences =
+      grokToolBudgetResidual.split("Tool budget:").length - 1;
+    expect(budgetOccurrences).toBe(1);
   });
 
-  test("no row denies use_skill — brief-named skills always load directly", () => {
-    for (const row of [defaultRow, museRow, grokRow]) {
-      expect(row.advertisedToolDeny).not.toContain("use_skill");
-    }
+  test("default row carries no residual", () => {
+    expect(defaultRow.residual).toBe("");
   });
 
-  test("sectionOmit starts empty on every row", () => {
-    for (const row of [defaultRow, museRow, grokRow]) {
-      expect([...row.sectionOmit]).toEqual([]);
-    }
+  test("claude row is the single XML task_guidance block (CL-8309)", () => {
+    expect(claudeRow.residual).toContain("<task_guidance>");
+    expect(claudeRow.residual).toContain("</task_guidance>");
+    expect(claudeRow.residual).toContain("Follow the dispatch brief exactly");
+    expect(claudeRow.residual).toContain("Batch independent tool calls");
+  });
+
+  test("gpt row is the narrate-before-tools nudge (CL-8310)", () => {
+    expect(gptRow.residual).toContain("Narrate before tools (GPT worker):");
+    expect(gptRow.residual).toContain(
+      "one short line saying what you are doing",
+    );
+    expect(gptRow.residual).toContain("done-definition is met");
   });
 });
diff --git a/packages/prompt-variance/src/rows.ts b/packages/prompt-variance/src/rows.ts
index dcac93a9d..d40979175 100644
--- a/packages/prompt-variance/src/rows.ts
+++ b/packages/prompt-variance/src/rows.ts
@@ -1,54 +1,42 @@
 /**
  * Versioned model-family prompt variance (CL-8269). One row per tuned
- * family: a tail residual plus the tool names the family must not
- * advertise. Directors assemble prompts through `assemble`, never by
- * hand-concatenating family booleans.
+ * family: the tail residual text directors append to the assembled prompt.
+ * Residuals-only: the package owns residual TEXT, never tool mounting —
+ * tool denial stays live in ModelFamilyPolicy.advertisedToolDeny
+ * (src/agent/model-family-policy.ts), which run.ts applies at mount time.
+ * Keeping deny out of this package removes the duplicate-deny footgun.
  *
- * Only families with shipped eval numbers live here: default/muse/grok.
- * glm/claude/gpt rows land when CL-8265 / 7772 / 7775 characterize them.
+ * Families ship here as their lanes characterize them: default/muse/grok
+ * first, claude (CL-8309) and gpt (CL-8310) folded in on the P5 rebase.
+ * Render position is always the tail so residuals cannot disturb the
+ * cached prompt prefix.
  */
 
 /** Families with a shipped variance row. */
-export type PromptVarianceFamily = "default" | "muse" | "grok";
-
-/** Residuals render at the tail so they cannot disturb the cached prefix. */
-export type PromptVarianceRender = "tail";
-
-export interface PromptVarianceRowOverride {
-  residual?: string;
-  advertisedToolDeny?: readonly string[];
-  sectionOmit?: readonly string[];
-}
+export type PromptVarianceFamily =
+  | "default"
+  | "muse"
+  | "grok"
+  | "claude"
+  | "gpt";
 
 export interface PromptVarianceRow {
   id: PromptVarianceFamily;
-  render: PromptVarianceRender;
   /** Tail text appended to the assembled prompt. Empty when the family needs none. */
   residual: string;
-  /**
-   * Tool names to drop from the advertised set (CL-7668). Subtractive
-   * only — a row can never grant a tool `assemble` was not given.
-   * `use_skill` is never denied: brief-named skills load directly.
-   */
-  advertisedToolDeny: readonly string[];
-  /** Assembly section ids to omit for this family. Empty until a family needs one. */
-  sectionOmit: readonly string[];
-  /** Optional per-model-id refinements, keyed by lowercase model id. */
-  overrides?: Record<string, PromptVarianceRowOverride>;
 }
 
 export const FAMILY_IDS: readonly PromptVarianceFamily[] = [
   "default",
   "muse",
   "grok",
+  "claude",
+  "gpt",
 ];
 
 export const defaultRow: PromptVarianceRow = {
   id: "default",
-  render: "tail",
   residual: "",
-  advertisedToolDeny: [],
-  sectionOmit: [],
 };
 
 // Muse Spark does not reliably stop a tool loop at medium reasoning effort:
@@ -57,38 +45,79 @@ export const defaultRow: PromptVarianceRow = {
 // size table and the constructor append cannot drift apart.
 export const museRow: PromptVarianceRow = {
   id: "muse",
-  render: "tail",
   residual:
     "Tool discipline:\n" +
     "- Batch independent tool calls into a single turn.\n" +
     "- Never re-read a file you have already read this session.\n" +
     "- Do not narrate; act.",
-  advertisedToolDeny: [],
-  sectionOmit: [],
 };
 
-// Tiny residual for Grok/xAI workers: mining showed higher tools-only thrash
-// than Codex on the same harness. Shared thrash harness + spawn contracts do
-// the structural work; this is only a finish-bias nudge, not a full rewrite.
-// Built from the current grok finish-bias text on origin/main; the CL-8297
-// residual folds in here at merge time.
+// Single grok finish-bias + ceremony residual (CL-8296): the finish-bias
+// bullets plus the three ceremony lines from the CL-7768 design (no git, no
+// pre-plan, verify once), merged into one block with no line twice. The
+// don't re-read idea appears exactly once (the "re-open paths" bullet).
+// Grok-only: detectModelFamily has no glm family, so no GLM row ships here.
 export const grokRow: PromptVarianceRow = {
   id: "grok",
-  render: "tail",
   residual: [
     "Finish bias (xAI / Grok worker):",
     "- Once you can answer the dispatch brief, prefer the structured report over another speculative tool call.",
     "- If the next call would only re-open paths you already read, write the report instead.",
     "- When the dispatch brief's done-definition is met, write the report envelope instead of making one more search or micro-edit.",
     "- Route file and web work through the dedicated tools, never run_shell — mining showed grok reaching for shell first when a typed tool already covered the job.",
+    "- Never run git add, git commit, git stash, or any other state-changing git command unless the user asks.",
+    "- Do not narrate a plan before acting on a small task; act, then report.",
+    "- Verify with the test command once at the end, not after every edit.",
+  ].join("\n"),
+};
+
+// Grok tool-budget hook (CL-8297): pure tool-loop budget, deliberately free
+// of ceremony lines and family-specific routing. Grok leaves carry this via
+// the ModelFamilyPolicy.promptResidual field alongside the grokRow block
+// above (the finish-bias note) — each once. A named export rather than a
+// family row because it travels a different seam (promptResidual field,
+// not the finish-bias note).
+export const grokToolBudgetResidual: string =
+  "Tool budget:\n" +
+  "- Batch independent tool calls into a single turn.\n" +
+  "- Never re-issue a tool call whose result you already have.\n" +
+  "- When the next call would only repeat prior work, write the report instead.";
+
+// Single XML residual for Claude-family workers (CL-8309): a prose residual
+// did nothing, but one <task_guidance> block cut Sonnet tokens. The block is
+// the whole residual — never a full-prompt XML renderer, never applied
+// outside the claude family. Rebuilt end to end from Anthropic's prompting
+// docs: rationale first, numbered approach, named output contract.
+export const claudeRow: PromptVarianceRow = {
+  id: "claude",
+  residual: [
+    "<task_guidance>",
+    "Autonomous coding turn: finish the work in this turn on your best judgment.",
+    "1. Follow the dispatch brief exactly; its Success criteria are the done-definition.",
+    "2. Batch independent tool calls into a single turn; work from files already read this session.",
+    "3. Finish the task when the done-definition is met: prefer the structured report envelope over another tool call.",
+    "</task_guidance>",
+  ].join("\n"),
+};
+
+// Tiny narrate-before-tools residual for GPT workers (CL-8310): GPT-5.5 runs
+// showed 6–13 silent tool-only turns. Shared thrash harness + spawn contracts
+// do the structural work; this is only a narrate-before-tools nudge.
+// Deliberately not manage_tasks ceremony — that is CL-7769, not this text.
+export const gptRow: PromptVarianceRow = {
+  id: "gpt",
+  residual: [
+    "Narrate before tools (GPT worker):",
+    "- Before each tool call, write one short line saying what you are doing and why.",
+    "- Never make back-to-back tool calls with no narration between them.",
+    "- When the dispatch brief's done-definition is met, write the report envelope instead of making another tool call.",
   ].join("\n"),
-  // Leaf value; resolvePromptVariance clears it for orchestrators.
-  advertisedToolDeny: ["skill_search"],
-  sectionOmit: [],
 };
 
 export const FAMILY_ROWS: Record<PromptVarianceFamily, PromptVarianceRow> = {
   default: defaultRow,
   muse: museRow,
   grok: grokRow,
+  claude: claudeRow,
+  gpt: gptRow,
 };
diff --git a/packages/prompt-variance/src/sizes.test.ts b/packages/prompt-variance/src/sizes.test.ts
index 66e2c618a..b8a54cb54 100644
--- a/packages/prompt-variance/src/sizes.test.ts
+++ b/packages/prompt-variance/src/sizes.test.ts
@@ -1,9 +1,5 @@
 import { describe, expect, test } from "bun:test";
-import { defaultRow, grokRow, museRow } from "./rows.js";
-
-// CL-8269 RED: per-family-row prompt sizes. Residuals are tail appends, so
-// each must stay a small fraction of a worker prompt: non-empty for the
-// tuned families, empty for default, and bounded well under a kilobyte each.
+import { claudeRow, defaultRow, gptRow, grokRow, museRow } from "./rows.js";
 
 describe("prompt-variance prompt sizes per family row", () => {
   test("default residual is empty", () => {
@@ -11,7 +7,7 @@ describe("prompt-variance prompt sizes per family row", () => {
   });
 
   test("every tuned residual stays under 2000 chars / 3000 bytes", () => {
-    for (const row of [museRow, grokRow]) {
+    for (const row of [museRow, grokRow, claudeRow, gptRow]) {
       expect(row.residual.length).toBeGreaterThan(0);
       expect(row.residual.length).toBeLessThan(2000);
       expect(Buffer.byteLength(row.residual, "utf8")).toBeLessThan(3000);
diff --git a/scripts/dead-export-allowlist.txt b/scripts/dead-export-allowlist.txt
index 3f3ddf02f..34948e783 100644
--- a/scripts/dead-export-allowlist.txt
+++ b/scripts/dead-export-allowlist.txt
@@ -51,19 +51,17 @@ packages/opencode-go/src/models.ts: satisfies
 packages/opencode-go/src/models.ts: readonly
 
 # Versioned prompt-variance package public surface (CL-8269 lane,
-# agent-owned). Family rows data, the assemble shape, and per-model-id
-# overrides re-exported through the package entry point for the 0.3.31+
-# agent packages; no in-repo barrel consumer exists yet (src/ consumes
-# assemble, resolvePromptVariance, grokRow, museRow, and PromptVarianceRow
-# only, and the package tests import the backing modules directly). Scoped
+# agent-owned). Family rows data re-exported through the package entry point;
+# no in-repo barrel consumer exists yet for the entries below (src/ consumes
+# claudeRow, gptRow, grokRow, grokToolBudgetResidual, and museRow through the
+# barrel, and the package tests import the backing modules directly). Scoped
 # to the exact ts-prune flags so a new dead export in this barrel fails the
 # guard instead of hiding under a whole-file exemption. Remove each entry
 # with the export it names if the public surface shrinks.
-packages/prompt-variance/src/index.ts: AssembledPromptVariance
-packages/prompt-variance/src/index.ts: applyRowOverride
+packages/prompt-variance/src/index.ts: assemble
+packages/prompt-variance/src/index.ts: resolvePromptVariance
 packages/prompt-variance/src/index.ts: defaultRow
 packages/prompt-variance/src/index.ts: FAMILY_IDS
 packages/prompt-variance/src/index.ts: FAMILY_ROWS
 packages/prompt-variance/src/index.ts: PromptVarianceFamily
-packages/prompt-variance/src/index.ts: PromptVarianceRender
-packages/prompt-variance/src/index.ts: PromptVarianceRowOverride
+packages/prompt-variance/src/index.ts: PromptVarianceRow
diff --git a/src/agent/model-family-policy.ts b/src/agent/model-family-policy.ts
index 721887b33..793c6f6dd 100644
--- a/src/agent/model-family-policy.ts
+++ b/src/agent/model-family-policy.ts
@@ -2,7 +2,13 @@ import {
   detectModelFamily,
   type ModelFamily,
 } from "../subagent/provider-family.js";
-import { museRow } from "../../packages/prompt-variance/src/index.js";
+import {
+  claudeRow,
+  gptRow,
+  grokRow,
+  grokToolBudgetResidual,
+  museRow,
+} from "../../packages/prompt-variance/src/index.js";
 
 /**
  * Per-model-family tuning for the shared directors (main chat director and
@@ -80,12 +86,9 @@ const DEFAULT_POLICY: Omit<ModelFamilyPolicy, "family"> = {
 // Generic 4-line tool-budget residual (CL-8297). Grok leaves get this via
 // promptResidual today; other families leave the seam unfilled until their
 // own lanes land. Deliberately free of ceremony lines and family-specific
-// routing — pure tool-loop budget.
-export const GROK_TOOL_BUDGET_RESIDUAL =
-  "Tool budget:\n" +
-  "- Batch independent tool calls into a single turn.\n" +
-  "- Never re-issue a tool call whose result you already have.\n" +
-  "- When the next call would only repeat prior work, write the report instead.";
+// routing — pure tool-loop budget. Single-sourced from the versioned
+// prompt-variance package (CL-8269); the name stays for existing importers.
+export const GROK_TOOL_BUDGET_RESIDUAL = grokToolBudgetResidual;
 
 // A directly observed 14-turn pure-tool-call session for this family
 // previously motivated a tightened nudge/pause pair here (6/10). That pair
@@ -139,29 +142,17 @@ const MUSE_POLICY: Omit<ModelFamilyPolicy, "family"> = {
 // prompt carries exactly one copy. This is a different block from the CL-8297
 // tool-budget hook (GROK_TOOL_BUDGET_RESIDUAL, surfaced via the
 // ModelFamilyPolicy.promptResidual field): grok leaves carry both, each once.
-export const GROK_PROMPT_RESIDUAL = [
-  "Finish bias (xAI / Grok worker):",
-  "- Once you can answer the dispatch brief, prefer the structured report over another speculative tool call.",
-  "- If the next call would only re-open paths you already read, write the report instead.",
-  "- When the dispatch brief's done-definition is met, write the report envelope instead of making one more search or micro-edit.",
-  "- Route file and web work through the dedicated tools, never run_shell — mining showed grok reaching for shell first when a typed tool already covered the job.",
-  "- Never run git add, git commit, git stash, or any other state-changing git command unless the user asks.",
-  "- Do not narrate a plan before acting on a small task; act, then report.",
-  "- Verify with the test command once at the end, not after every edit.",
-].join("\n");
+// Single-sourced from the versioned prompt-variance package (CL-8269); the
+// name stays for existing importers.
+export const GROK_PROMPT_RESIDUAL = grokRow.residual;
 // Claude (Anthropic) ships one XML residual, not prose: a prose residual did
 // nothing, but a single <task_guidance> block cut Sonnet tokens. The block is
 // the whole residual — never a full-prompt XML renderer. The text lives here
 // (policy owns data); buildClaudeTaskGuidanceNote (prompts.ts) returns it
-// verbatim so the prompt carries exactly one copy.
-export const CLAUDE_TASK_GUIDANCE_NOTE = [
-  "<task_guidance>",
-  "Autonomous coding turn: finish the work in this turn on your best judgment.",
-  "1. Follow the dispatch brief exactly; its Success criteria are the done-definition.",
-  "2. Batch independent tool calls into a single turn; work from files already read this session.",
-  "3. Finish the task when the done-definition is met: prefer the structured report envelope over another tool call.",
-  "</task_guidance>",
-].join("\n");
+// verbatim so the prompt carries exactly one copy. Single-sourced from the
+// versioned prompt-variance package (CL-8269); the name stays for existing
+// importers.
+export const CLAUDE_TASK_GUIDANCE_NOTE = claudeRow.residual;
 
 const CLAUDE_POLICY: Omit<ModelFamilyPolicy, "family"> = {
   ...DEFAULT_POLICY,
@@ -175,13 +166,9 @@ const CLAUDE_POLICY: Omit<ModelFamilyPolicy, "family"> = {
 // The text lives here (policy owns data); buildGptNarrateBeforeToolsNote
 // (prompts.ts) returns it verbatim so the prompt carries exactly one copy.
 // Served cells (astra/sol/terra/…) are never named here — CL-8265
-// characterizes them later.
-export const GPT_NARRATE_BEFORE_TOOLS_NOTE = [
-  "Narrate before tools (GPT worker):",
-  "- Before each tool call, write one short line saying what you are doing and why.",
-  "- Never make back-to-back tool calls with no narration between them.",
-  "- When the dispatch brief's done-definition is met, write the report envelope instead of making another tool call.",
-].join("\n");
+// characterizes them later. Single-sourced from the versioned
+// prompt-variance package (CL-8269); the name stays for existing importers.
+export const GPT_NARRATE_BEFORE_TOOLS_NOTE = gptRow.residual;
 
 // GPT (Codex / gpt-*) thresholds are provisional: we have no eval
 // characterization yet for how GPT behaves under tool-only stretches or
diff --git a/src/agent/prompt-sizes.test.ts b/src/agent/prompt-sizes.test.ts
index 5c2985f1d..dcc9335d9 100644
--- a/src/agent/prompt-sizes.test.ts
+++ b/src/agent/prompt-sizes.test.ts
@@ -127,10 +127,16 @@ function budgetMessage(
 describe("director prompt size budget", () => {
   const rows = directorPromptSizeTable();
 
-  test("covers every director in all three variance families", () => {
-    expect(rows.length).toBe(DIRECTOR_IDS.length * 3);
+  test("covers every director in all five variance families", () => {
+    expect(rows.length).toBe(DIRECTOR_IDS.length * 5);
     for (const directorId of DIRECTOR_IDS) {
-      for (const family of ["default", "muse", "grok"] as const) {
+      for (const family of [
+        "default",
+        "muse",
+        "grok",
+        "claude",
+        "gpt",
+      ] as const) {
         expect(
           rows.some((r) => r.directorId === directorId && r.family === family),
         ).toBe(true);
@@ -192,6 +198,38 @@ describe("director prompt size budget", () => {
     }
   });
 
+  test("claude leaves carry the XML task_guidance block exactly once", () => {
+    for (const directorId of DIRECTOR_IDS) {
+      const prompt = assembleDirectorPrompt(directorId, "claude");
+      const occurrences = prompt.split("<task_guidance>").length - 1;
+      if (DIRECTOR_REGISTRY[directorId].spawn.maySpawn) {
+        expect(occurrences).toBe(0);
+      } else {
+        expect(occurrences).toBe(1);
+        expect(prompt).toContain("</task_guidance>");
+      }
+    }
+  });
+
+  test("gpt directors carry the narrate-before-tools nudge exactly once", () => {
+    for (const directorId of DIRECTOR_IDS) {
+      const prompt = assembleDirectorPrompt(directorId, "gpt");
+      const occurrences =
+        prompt.split("Narrate before tools (GPT worker):").length - 1;
+      expect(occurrences).toBe(1);
+    }
+  });
+
+  test("default family carries no residual", () => {
+    for (const directorId of DIRECTOR_IDS) {
+      const prompt = assembleDirectorPrompt(directorId, "default");
+      expect(prompt).not.toContain("<task_guidance>");
+      expect(prompt).not.toContain("Narrate before tools (GPT worker):");
+      expect(prompt).not.toContain("Tool budget:");
+      expect(prompt).not.toContain("Finish bias (xAI / Grok worker):");
+    }
+  });
+
   test("measurement is deterministic", () => {
     const again = directorPromptSizeTable();
     expect(again.map((r) => r.chars)).toEqual(rows.map((r) => r.chars));
@@ -200,7 +238,13 @@ describe("director prompt size budget", () => {
 
   test("tool names match the production mount: no dupes, no phantoms", () => {
     for (const directorId of DIRECTOR_IDS) {
-      for (const family of ["default", "muse", "grok"] as const) {
+      for (const family of [
+        "default",
+        "muse",
+        "grok",
+        "claude",
+        "gpt",
+      ] as const) {
         const names = canonicalToolNamesForDirector(
           DIRECTOR_REGISTRY[directorId],
           family,
@@ -234,7 +278,7 @@ describe("director prompt size budget", () => {
   test("formatPromptSizeTable renders one row per director", () => {
     const table = formatPromptSizeTable(rows);
     expect(table).toContain(
-      "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) |",
+      "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) | claude chars (bytes) | gpt chars (bytes) |",
     );
     for (const directorId of DIRECTOR_IDS) {
       const base = rows.find(
@@ -246,8 +290,14 @@ describe("director prompt size budget", () => {
       const grok = rows.find(
         (r) => r.directorId === directorId && r.family === "grok",
       );
+      const claude = rows.find(
+        (r) => r.directorId === directorId && r.family === "claude",
+      );
+      const gpt = rows.find(
+        (r) => r.directorId === directorId && r.family === "gpt",
+      );
       expect(table).toContain(
-        `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
+        `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) | ${claude?.chars} (${claude?.bytes}) | ${gpt?.chars} (${gpt?.bytes}) |`,
       );
     }
   });
diff --git a/src/agent/prompt-sizes.ts b/src/agent/prompt-sizes.ts
index b6ce09f46..54b55b12b 100644
--- a/src/agent/prompt-sizes.ts
+++ b/src/agent/prompt-sizes.ts
@@ -58,9 +58,9 @@ export const CANONICAL_PROMPT_ENV: EnvironmentInfo = {
 };
 
 const GROK_PROVIDER = { providerName: "xai/default", model: "grok-4.6" };
-// Default-family probe: an unrecognized provider stays on the default
-// family no matter how many family rows land (claude/gpt already ship),
-// so the default column carries no residual.
+// Default-family probe: anthropic/claude-sonnet-4 hits the claude row and
+// openai/gpt-4.1 hits the gpt row, so an unrecognized provider is the probe
+// that still resolves to the default family (no residual).
 const MUSE_PROVIDER = {
   providerName: "opencode-go",
   model: "muse-spark-1.3-contributor",
@@ -69,9 +69,14 @@ const DEFAULT_PROVIDER = {
   providerName: "unknown-provider",
   model: "unknown-model",
 };
+const CLAUDE_PROVIDER = {
+  providerName: "anthropic",
+  model: "claude-sonnet-4",
+};
+const GPT_PROVIDER = { providerName: "openai", model: "gpt-4.1" };
 
-/** Families in the size table: default, muse (+tool-discipline rules), grok (+finish-bias note). */
-export type PromptSizeFamily = "default" | "muse" | "grok";
+/** Families in the size table: default (no residual), muse, grok, claude, gpt. */
+export type PromptSizeFamily = "default" | "muse" | "grok" | "claude" | "gpt";
 
 /**
  * Pinned AGENTS.md body for prefix measurement. Production reads the live
@@ -125,7 +130,13 @@ export function canonicalToolNamesForDirector(
   const providerName =
     family === "grok"
       ? GROK_PROVIDER.providerName
-      : DEFAULT_PROVIDER.providerName;
+      : family === "muse"
+        ? MUSE_PROVIDER.providerName
+        : family === "claude"
+          ? CLAUDE_PROVIDER.providerName
+          : family === "gpt"
+            ? GPT_PROVIDER.providerName
+            : DEFAULT_PROVIDER.providerName;
   const filtered = [...preFilterMountNames(isCodexProviderName(providerName))];
   const capabilities = packageToCapabilities(pkg);
   const names =
@@ -172,9 +183,13 @@ export function assembleDirectorPrompt(
       ? GROK_PROVIDER
       : family === "muse"
         ? MUSE_PROVIDER
-        : DEFAULT_PROVIDER;
+        : family === "claude"
+          ? CLAUDE_PROVIDER
+          : family === "gpt"
+            ? GPT_PROVIDER
+            : DEFAULT_PROVIDER;
   const policy = resolveModelFamilyPolicy({ ...provider, orchestrator });
-  return buildSubAgentSystemPrompt(
+  const prompt = buildSubAgentSystemPrompt(
     [formatDirectorSystemPrompt(pkg)],
     CANONICAL_PROMPT_ENV,
     undefined,
@@ -185,6 +200,12 @@ export function assembleDirectorPrompt(
       promptResidual: policy.promptResidual,
     },
   );
+  // Mirror the SubAgentDirector constructor (nudge-director.ts): family
+  // tool-discipline rules go at the tail of the prompt on the wire.
+  return policy.toolDisciplineRules !== undefined &&
+    policy.toolDisciplineRules.length > 0
+    ? `${prompt}\n\n${policy.toolDisciplineRules}`
+    : prompt;
 }
 
 /**
@@ -244,7 +265,13 @@ export function measureSkywalkerPrefix(): SkywalkerPrefixSize {
 export function directorPromptSizeTable(): DirectorPromptSize[] {
   const rows: DirectorPromptSize[] = [];
   for (const directorId of DIRECTOR_IDS) {
-    for (const family of ["default", "muse", "grok"] as const) {
+    for (const family of [
+      "default",
+      "muse",
+      "grok",
+      "claude",
+      "gpt",
+    ] as const) {
       rows.push(measureDirectorPrompt(directorId, family));
     }
   }
@@ -254,8 +281,8 @@ export function directorPromptSizeTable(): DirectorPromptSize[] {
 /** Render the size table as markdown (for PR bodies and budget updates). */
 export function formatPromptSizeTable(rows: DirectorPromptSize[]): string {
   const lines = [
-    "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) |",
-    "| --- | --- | --- | --- |",
+    "| director | default chars (bytes) | muse chars (bytes) | grok chars (bytes) | claude chars (bytes) | gpt chars (bytes) |",
+    "| --- | --- | --- | --- | --- | --- |",
   ];
   for (const directorId of DIRECTOR_IDS) {
     const base = rows.find(
@@ -267,8 +294,14 @@ export function formatPromptSizeTable(rows: DirectorPromptSize[]): string {
     const grok = rows.find(
       (r) => r.directorId === directorId && r.family === "grok",
     );
+    const claude = rows.find(
+      (r) => r.directorId === directorId && r.family === "claude",
+    );
+    const gpt = rows.find(
+      (r) => r.directorId === directorId && r.family === "gpt",
+    );
     lines.push(
-      `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) |`,
+      `| ${directorId} | ${base?.chars} (${base?.bytes}) | ${muse?.chars} (${muse?.bytes}) | ${grok?.chars} (${grok?.bytes}) | ${claude?.chars} (${claude?.bytes}) | ${gpt?.chars} (${gpt?.bytes}) |`,
     );
   }
   return lines.join("\n");